Zika virus-induced acute myelitis and motor deficits in adultinterferon αβ/γ receptor knockout mice

Katherine Zukor1 & Hong Wang1 & Venkatraman Siddharthan1& Justin G. Julander1 &

John D. Morrey1

Received: 31 August 2017 /Revised: 11 October 2017 /Accepted: 23 October 2017 /Published online: 23 February 2018# The Author(s) 2018. This article is an open access publication

Abstract Zika virus (ZIKV) has received widespread at-tention because of its effect on the developing fetus. It isbecoming apparent, however, that severe neurological se-quelae, such as Guillian-Barrë syndrome (GBS), myelitis,encephalitis, and seizures can occur after infection ofadults. This study demonstrates that a contemporarystrain of ZIKV can widely infect astrocytes and neuronsin the brain and spinal cord of adult, interferon α/βreceptor knockout mice (AG129 strain) and cause pro-gressive hindlimb paralysis, as well as severe seizure-like activity during the acute phase of disease. The se-verity of hindlimb motor deficits correlated with in-creased numbers of ZIKV-infected lumbosacral spinalmotor neurons and decreased numbers of spinal motorneurons. Electrophysiological compound muscle actionpotential (CMAP) amplitudes in response to stimulationof the lumbosacral spinal cord were reduced when obvi-ous motor deficits were present. ZIKV immunoreactivitywas high, intense, and obvious in tissue sections of thebrain and spinal cord. Infection in the brain and spinalcord was also associated with astrogliosis as well as Tcell and neutrophil infiltration. CMAP and histologicalanalysis indicated that peripheral nerve and muscle func-

tions were intact. Consequently, motor deficits in thesecircumstances appear to be primarily due to myelitis andpossibly encephalitis as opposed to a peripheral neurop-athy or a GBS-like syndrome. Thus, acute ZIKV infec-tion of adult AG129 mice may be a useful model forZIKV-induced myelitis, encephalitis, and seizure activity.

Keywords Zika virus . Brain . Spinal cord .Motor .

Guillian-Barrë syndrome .Mouse

AbbreviationsZIKV Zika virusGBS Guillain-Barré syndromeA129 and IFNAR−/−

miceInterferon type 1 (αβ)receptor knockout mice

AG129 Interferon types 1 and 2 (αβ/γ) receptorknockout mice

DPI Days post-infectionCMAP Compound muscle action potentialVPS Viral paresis scalePBS Phosphate buffered salineNMJ Neuromuscular junctionNF-H Neurofilament-HTMR Tetramethylrhodamineα-btx α-BungarotoxinPBST PBS with 0.5% Triton

X-100ROI Region of interestChAT Choline acetyltransferaseMBP Myelin basic proteinir Immunoreactive

Electronic supplementary material The online version of this article(https://doi.org/10.1007/s13365-017-0595-z) contains supplementarymaterial, which is available to authorized users.

* John D. Morreyjohn.morrey@usu.edu

1 Institute for Antiviral Research, Department of Animal, Dairy, andVeterinary Sciences, Utah State University, 5600 Old Main Hill,Logan, UT 84322-5600, USA

J. Neurovirol. (2018) 24:273–290https://doi.org/10.1007/s13365-017-0595-z

Introduction

Zika virus (ZIKV) is an emerging flavivirus that has receivedwidespread attention because of its effect on the developingfetus. In utero infections cause congenital defects, most nota-bly microcephaly along with other deformities (Schuler-Faccini et al. 2016). While ZIKV infection of adults generallyproduces only a mild disease, it is becoming apparent that, aswith other flavivirus infections, severe neurological sequelaecan occur. Recent outbreaks have been associated with higherincidences of peripheral neuropathies such as Guillian-Barrësyndrome (GBS) (Brasil et al. 2016; Cao-Lormeau et al. 2016;Cardoso et al. 2015; Samarasekera and Triunfol 2016) andother neurological diseases such as myelitis (Anaya et al.2017; Dirlikov et al. 2016;Mecharles et al. 2016), encephalitis(Carteaux et al. 2016; Nicastri et al. 2016; Soares et al. 2016),seizures (Asadi-Pooya 2016), and various ophthalmologicalconditions (Pastula et al. 2016; Smith et al. 2016). This isnot surprising considering that related flaviviruses such asWest Nile virus (WNV) and Japanese encephalitis virus(Solomon et al. 1998) can cause myelitis and motor deficits(Sejvar et al. 2003). Given the emerging nature of ZIKV, how-ever, it is likely that we do not fully understand the acute andlong-term consequences of ZIKV infection on the nervoussystem. Therefore, investigating the pathobiology of ZIKVin animal models will help to understand the potential forZIKV to cause neurological disease in human subjects. Wefocus herein on adult models because ZIKV-related motordeficits have been primarily associated with adult infection,as opposed to in utero infection (Anaya et al. 2017; Dirlikovet al. 2016; Mecharles et al. 2016).

Prior to the 2015 ZIKVoutbreak, a few animal models ofZIKV infection existed, but infection of rodents was largelynon-productive unless a lab strain of the virus that had un-dergone several serial passages in mice was used (Dick1952). After the recent outbreaks, efforts to develop animalmodels with more clinically relevant isolates of the virushave been renewed. A rapid series of publications found thatadult mice that lack type 1 (αβ; A129 and IFNAR−/− strains)or types 1 and 2 (αβ/γ; AG129 strain) interferon receptorsare susceptible to lethal infection (Aliota et al. 2016; Lazearet al. 2016; Manangeeswaran et al. 2016; Rossi et al. 2016;Zmurko et al. 2016). These models may have some relevanceto human ZIKV infections in that, like many viruses, ZIKVgains advantages in human hosts by inhibiting interferonresponses (Best 2017; Bowen et al. 2017; Schulz andMossman 2016). Because viruses may not be able to inhibitmouse-specific interferon pathways (Aguirre et al. 2012),blocking them by other means, as in AG129 mice, may moreclosely mimic what happens in humans. Although AG129mice are deficient in innate immune responses, they do elicitacquired immune responses, such as vaccine-elicited protec-tive immunity (Sumathy et al. 2017; Weger-Lucarelli et al.

2017). While complete knockout of the interferon responsesproduces a more severe disease in mice, it can provide in-sights into what happens when ZIKV gains access to theadult central nervous system (CNS). Besides a lethal infec-tion, these mice manifest neurological symptoms such asBtoe walking,^ tremors, loss of balance, paralysis, andhunched posture (Aliota et al. 2016; Lazear et al. 2016;Manangeeswaran et al. 2016; Rossi et al. 2016).

To better understand the consequences of ZIKV infectionof the adult nervous system and the pathobiology of theresulting neurological disease, we infected AG129 mice witha contemporary, low-passage ZIKV isolate and evaluated theneurological motor deficits occurring during the acute phaseof the disease.We characterized the onset, course, and severityof behavioral hindlimb deficits and used electrophysiologyand immunohistochemistry (IHC) to determine if such deficitsare due primarily to myelitis, peripheral neuropathy, myositis,or encephalitis. This included histological analysis of motorcortex, spinal cord, peripheral nerve, and muscle.

Materials and methods

Animals

Male and female AG129 mice (van den Broek et al. 1995)were bred in-house in sterilized cages and maintained in a 12/12 light cycle. Mice were randomly assigned to treatmentgroups based on weight, gender, and baseline measurements.

Virus

A Puerto Rican isolate of ZIKV (PRVABC59, Human/2015/Puerto Rico, GenBank KU501215) was obtained from BEIResources (Cat No. NR-50240, Lot No. 64112564). The cer-tificate of analysis confirmed that the sequence of this stock(GenBank KX087101) is 99% identical to PRVABC59(GenBank KU501215). The BEI stock was passaged twotimes in Vero 76 cells to make a stock with a titer of2 × 107 pfu/mL for use in all experiments. Infected cells werefrozen once, thawed, centrifuged to remove cell debris, andaliquoted in frozen stocks. Dilutions were made in minimalessential medium supplemented with 50 μg/mL gentamicin todeliver 2000 pfu subcutaneously in the inguinal area on theright side in a volume of 0.1 mL. Uninfected cells were pre-pared and diluted similarly for sham infections.

Experiment no. 1

The purpose of this experiment was to assess the time courseand severity of motor deficits and collect tissues for histolog-ical analysis. Seventy-two to 75-day-old (10.5 weeks) maleand female AG129 mice were infected with ZIKV (n = 6

274 J. Neurovirol. (2018) 24:273–290

females and 4 males) or sham (n = 3 females and 2 males)inoculum and monitored for weight loss and survival (Fig. 1).Behavioral motor assessments were performed before infec-tion (baseline); once after infection, but before symptom onset(4 days post-infection (DPI)); and twice a day after symptomonset (9–15 DPI) (Fig. 2). Seizure-like activity observed dur-ing behavioral assessments was also noted (Fig. 3). Videos ofmotor deficits and seizure-like activity were obtained to pro-vide examples of symptoms seen. Observations were made byresearchers who were blind to the infection status of eachgroup. Moribund mice were perfused for histological analysisof the brain, lumbosacral spinal cord, sciatic nerve, and gas-trocnemius muscle.

Experiment no. 2

The purpose of this experiment was to collect electrophysio-logical data on mice with motor deficits. Male and femaleAG129 mice aged 120–123 days old (17 weeks) or 62 daysold (9 weeks) were divided into groups. The ZIKV-infectedgroup contained three males (17 weeks old), four males(9 weeks old), and four females (9 weeks old). The sham-infected group contained twomales (17 weeks old), twomales(9 weeks old), and two females (9 weeks old). Mice weremonitored for weight loss and survival. Behavioral assess-ments were performed before infection (baseline) and aftersymptom onset so that mice with moderate to severe deficits(scores 3–6, see below) could undergo electrophysiologicalassessment. Compound muscle action potential (CMAP) ofthe gastrocnemius muscle was recorded before infection(baseline) and after moderate to severe deficits had developed.After infection, one sham-infected mouse was recorded forevery one to two ZIKV-infected mouse. Measurements wereobtained by a researcher who was blind to the infection statusand deficit score of each mouse. CMAP responses were re-corded in response to stimulation at the sciatic notch and then

the lumbosacral spinal cord. Because stimulation of the lum-bar spinal cord was an invasive procedure, we did not obtainbaseline measurements for CMAP in response to spinal cordstimulation.

Viral paresis scale

Mice were analyzed for signs of tail and hindlimb paresis/paralysis using a sensitive, open-field assay modified fromthe Basso Mouse Scale used to assess paralysis in spinalcord-injured mice (Basso et al. 2006) and a test used to trackparalysis in amyotrophic lateral sclerosis mouse models(Hatzipetros et al. 2015). Eachmouse was placed on a tabletopand allowed to roam freely for 4 min. Hindlimb function wasscored on a seven-point scale detailed in Table 1 by re-searchers who were blind to the infection status of each group.Scoring was based on four main categories: tail position dur-ing walking, miss-step severity, weight bearing, and jointmovement. Miss-step severity was scored only on assessablewalking passes, which was defined as a pass in which theanimal moved three body lengths at a consistent speed andwithout turning (Basso et al. 2006). Separate scores were giv-en for the left and right hindlimbs to assess if symptoms werebilateral or unilateral.

Seizure-like activity score

Seizure-like activity noted during viral paresis scale (VPS)assessments was recorded as mild if the animal had mildshakes or tremors or severe if the animal was violently seizingin the cage.

Histological analyses

Mice were perfused transcardially with phosphate-bufferedsaline (PBS) followed by 4% paraformaldehyde. The brain,

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

-30

-20

-10

0

10

days after ZIKV infection

% w

eigh

t cha

nge

0 5 10 15 200

20406080

100

days after ZIKV infection

% s

urvi

val

1A-7021A-7031A-7071B-709

1C-713

1B-7041B-705

1C-7101C-7111C-712

2A-7012A-7062A-708

2B-7142B-715

shamfemale sham

male

ZIKVmale

ZIKVfemalea

b

Fig. 1 Weight loss and survival curves for AG129 adult mice infectedwith ZIKV in experiment no. 1. a Body weight expressed as a percentageof starting weight after ZIKV or sham infection. b Percentage of each

group surviving at each day after ZIKV or sham infection. All sham-infected animals (male and female) are represented in the sham group

J. Neurovirol. (2018) 24:273–290 275

kidney, liver, pancreas, hindlimbs, and lumbosacral spinal col-umn were removed and post-fixed in the same fixative over-night, rocking at 4 °C. Tissues were rinsed twice in PBS, andthe lumbosacral spinal cord, sciatic nerves, and gastrocnemiusmuscles were isolated. All tissues except the gastrocnemiusmuscle on the right were cryoprotected in 30% sucrose in PBSfor 2–3 days at 4 °C, embedded in OCT compound (TedPella), and frozen with a dry ice/ethanol bath. Five sets ofadjacent sections were cut on a cryostat at 25 μm, mountedon slides, and stored at − 20 °C until ready for further process-ing. For immunohistofluorescent staining, sections wereencircled with a hydrophobic barrier pen (ImmEdge, VectorLabs), rinsed with PBS, and blocked with 10% normal serumand 1% Triton X-100 in PBS for 1 h. Primary antibodies were

diluted in blocking solution as shown in Table 2 and applied tosections for incubation overnight at room temperature.Secondary antibodies conjugated to Alexa-488, Alexa-568,or Alexa-647 (from Invitrogen or Jackson ImmunoResearch)were diluted to 10 μg/mL in blocking solution. Sections wererinsed three times in PBS, incubated with secondary antibodysolution for 2 h at room temperature, rinsed three times inPBS, incubated with Hoechst 33342 (Invitrogen, 1/2000 inPBS with 0.05% Triton X-100), and rinsed twice in PBS.Coverslips were mounted with Fluoromount G (SouthernBiotech). For CD3 labeling, the amount of Triton X-100 wasdecreased to 0.5% in the blocking solution and 0.2% in theantibody diluent.

Gastrocnemius muscles from the right side were sectionedlongitudinally at 200 μm on a Vibratome (3000 plus,Vibratome Co.), collected in PBS, and stored at 4 °C. Ten to12 sections were obtained for each muscle, and the third orfourth and seventh or eighth sections were chosen for neuro-muscular junction (NMJ) and neurofilament (NF-H) stainingas described in Itoh et al. (2011). Briefly, sections were incu-bated free floating in 0.1M glycine in PBS, pH 7.3 for 30min;blocked with 5% donkey serum (Jackson ImmunoResearchLaboratories), 0.5% Triton X-100, and 1% BSA in PBS con-taining tetramethylrhodamine (TMR)-conjugated alpha-bungarotoxin (α-btx) for 1 h; permeablized with 100% meth-anol at − 20 °C for 7 min; rinsed with PBS with 0.5% TritonX-100 (PBST); and incubated with primary antibody dilutedin 1% BSA and 0.3% Triton X-100 in PBS at 4 °C for 48 h.After rinsing in PBST three times, sections were incubated atroom temperature for 2 h with secondary antibody (anti-

0 5 10 15

0

1

2

3

4

5

6

days after ZIKV infection

VP

S s

core

(left

side

)0 5 10 15

0

1

2

3

4

5

6

days after ZIKV infection

VP

S s

core

(rig

ht s

ide)

0 5 10 15

0

1

2

3

4

5

6

days after ZIKV infection

VP

S s

core

(rig

ht s

ide)

0 5 10 15

0

1

2

3

4

5

6

days after ZIKV infection

VP

S s

core

(left

side

)

shamZIKV-early deathZIKV-late death

ba

dc

1A-702

1A-703

1A-707

1B-704

1B-705

1B-709

1C-710

1C-711

1C-712

1C-713 2A-701

2A-706

2A-708

shamfemale

ZIKVmale

ZIKVfemale

Fig. 2 AG129 mice infected with ZIKV experience moderate paresis tosevere paralysis before death. a, b VPS scores from experiment no. 1 ofindividual mice on the right (a) and left sides (b). c, dMeanVPS scores of

ZIKV-infected mice that died early (n = 5) or late (n = 5) compared tosham-infected mice (n = 3) on the right (c) and left sides (d). Error barsrepresent SEM

0

2

4

6

seizure severity

high

est

VP

S s

core

none

mild

seve

re

ZIKV

sham

02468

10

# an

imal

s

mildnone

severe

seizureseverity

a b

Fig. 3 Seizure activity occurs independently of hindlimb motor deficitlevel. a Number of animals in sham-infected or ZKIV-infected groupsfrom experiment no. 1 having mild or severe seizure activity. bCorrelation between highest VPS score and seizure severity for ZIKV-infected animals. Each dot represents one animal. The line is the best fitline if 0, 1, and 2 are used for seizure categories of none, mild, and severe,respectively. Slope is not significantly different from zero

276 J. Neurovirol. (2018) 24:273–290

chicken conjugated to Alexa Fluor® 488, 1/100, JacksonImmunoResearch Laboratories) diluted in PBS. After rinsingin PBST three times, sections were dehydrated through amethanol series (15 min each in 20, 50, 70, 100%), clearedwith benzyl alcohol/benzyl benzoate (1:2) (Zukor et al. 2010),and stored at 4 °C until ready to image.

Imaging and image processing

Fluorescent images were obtained at ×10 or ×20 with a laserscanning confocal microscope (Zeiss, LSM710) equippedwith 405, 488, 561, and 633 laser lines. For images takenfor pixel-based quantification, identical settings were usedfor all images in a set. For images chosen for publication,distracting artifacts were removed in ImageJ (Schneideret al. 2012) and levels were adjusted in Photoshop to maxi-mize the signal-to-noise ratio so that relevant features could beseen more clearly. For images chosen to highlight pixel-basedquantification, sham and ZIKV group images were adjustedidentically to enable equitable comparison.

Image quantification

ImageJ or FIJI was used for image quantification (Schneideret al. 2012). The cell counter plugin was used for manual cellcounts. For pixel-based quantification of a given antibody

Table 1 VPS scoring criteria

Score Description Signs

0 Normal Normal, weight-bearing, plantar steppinga with tail up during walking passesb

1 Onset of symptoms Weight-bearing, plantar stepping with mild rotation

Tail position: May be down or not fully up

Miss-step: Foot rotated on take-off or landing

Weight bearing: Wobble is present indicating weakness

2 Mild paresis Mild miss-steps (but able to bear weight)

Tail position: Down

Miss-step: Mild, toe curling/dragging on ground foot slightly skids medially or laterally

Weight bearing: A limp may be present indicating weakness

Joint movement: May appear stiffer

3 Moderate paresis Moderate miss-steps (but able to bear weight)

Miss-step: Obvious foot curling/dragging on ground foot obviously skids medially or laterally

Weight bearing: Limb is obviously weak

Joint movement: May appear stiffer

4 Severe paresis Severe miss-steps (not bearing much weight)

Miss-step: Limb mostly drags behind, medially or laterally

Weight bearing: Not much, but limb still used to aid forward motion

Joint movement: Obviously decreased

5 Paralysis No weight-bearing steps, slight joint movement

Miss-step: No stepping, limb only drags

Weight bearing: None

Joint movement: Slight

6 Complete paralysis No weight-bearing steps, no joint movement

Joint movement: None

a Plantar stepping: Paw is placed flat on ground during stepping. It does not curl or skid to one side, and the toes/feet do not curl or drag at any pointbWalking pass: Animal moves three body lengths at a consistent speed and without turning

Table 2 Primary antibodiesa

Antibody Antibody type Company, Catalog # Dilution

ChAT Goat pAb Millipore, AB144P 1/100

ZIKV Rabbit pAb IBT, 0308-001 1/500

iba1 Goat pAb Abcam, ab5076 1/200

GFAP Rat IgG2a-kappa mAb Invitrogen, 13-0300 1/500

CD3 Rabbit mAb Abcam, ab16669 1/100

NF-H Chick pAb Aves Labs, NFH 1/200

MBP Rat IgG2a mAb Abcam, ab7349 1/200

Ly6G Rat IgG2b mAb Abcam, ab25377 1/500

α-Bungarotoxin-TMR Biotium, 00012/00014 1/1000

pAb polyclonal antibody, mAb monoclonal antibodya Summary of primary antibodies and dilutions used

J. Neurovirol. (2018) 24:273–290 277

signal, the area to be measured was defined by a region ofinterest (ROI); then, thresholds of single-channel images wereadjusted to select pixels with signal above noise (positivepixels). Thresholds of all images in a set were adjusted iden-tically for equitable comparison. Then, positive pixels withinthe whole area ROI were measured to obtain the area occupiedby positive pixels. The area of positive pixels was divided bythe whole area of the main ROI to determine what proportionof the ROI was positive for the antibody. For co-localizationanalysis, pixels that were positive for two antibody signalswere measured.

Sagittal sections of the brain containing motor cortex wereidentified based on the morphology of the lateral ventricleabout 1 mm lateral of bregma (Paxinos and Franklin 2004).A 1417 × 553 μm2 region containing the pyramidal cell layerof the motor cortex, in a region of the motor cortex just dorsalto the lateral ventricle, was measured for pixel-based quanti-fications (see box in Fig. 4 (D) for location of the regionmeasured). Two sections per animal were measured for eachparameter. For analysis of ZIKV in the hippocampus, onesection per animal was measured, and the area measuredcorresponded to pyramidal cell layer (see Fig. 4 (B) for loca-tion of region measured).

Spinal cord levels were identified using the mouse spinalcord atlas (Watson et al. 2009) and the location and morphol-ogy of clusters of choline acetyltransferase (ChAT) positiveneurons. Because motor neurons for the gastrocnemius mus-cle are located at the L4-L5 level (McHanwell and Biscoe1981), sections from this level were chosen for analysis whenpossible. For ventral motor neuron, astrocyte, and ZIKV in-fection level analyses, two sections per animal were analyzed.ZIKV−, ChAT+ and ZIKV+, ChAT+ neurons in the ventralhorns were counted manually using the cell counter in ImageJ.Because there were occasionally ZIKV+ cells in the ventralhorns with neuronal morphology that had low or undetectableChAT levels (Fig. 5 (B1–B3)), these cells were also counted asventral motor neurons. For analysis of Ly6G (neutrophils),CD3 (T cells), and iba1 (microglia/macrophages) signals,one section per animal was analyzed.

For the nerve cross sections, a stereological countinggrid containing 25 × 25 μm2 boxes where each box wasspaced 25 μm from the boxes above and below it wasplaced over the nerve to obtain counts from a sample ofthe nerve, representing approximately one fourth of thetotal area. An ROI was placed around the whole nerve,and the counting grid was cut to eliminate all areas of thegrid outside of the nerve (see Fig. 6a). Stereologicalcounting rules were used to count the number of myelinat-ed axons, unmyelinated axons, and empty myelin sheathswithin the boxes of the cut grid. Images were converted tocomposite images, so the neurofilament and myelin basicprotein (MBP) channels could be toggled on and off duringcounting. All counts were done by a researcher who was

blind to the status of each animal. The area of the cut gridas well as the area of the whole ROI were calculated andused to extrapolate the number of myelinated axons, un-myelinated axons, and empty myelin sheaths within thewhole area. Three sections per animal were analyzed.

For NMJ analysis, two sections per animal were analyzed.Confocal z-stacks (~ 1-μm step size) of two optimal fields ofview containing NMJ endplates (α-btx+) and nerve fibers(NF-H+) were analyzed per section for a total of four fieldsof view per animal. The number of endplates and the numberof innervated endplates (co-localized with someneurofilament signal) were counted manually with the cellcounter in ImageJ. Images were converted to composite im-ages, so the NF-H andα-btx channels could be toggled on andoff during counting. Total numbers of innervated NMJs (NF-H+, α-btx+) and all NMJs (α-btx+) from all four fields ana-lyzed were used to calculate the percentage of NMJs that wereinnervated per animal.

Gastrocnemius CMAP

CMAP measurements were performed by a researcher whowas blind to the infection status and VPS score of eachmouse. Animals were anesthetized with isoflurane and main-tained at 37 °C body temperature with a rectal probe andheating pad, and their hindlimbs were shaved. Monopolarneedle electrodes (Tai-Chi Brand, acupuncture needles) wereinserted into the sciatic notch or epidural to the lumbosacralspinal cord (after surgical exposure) to stimulate the gastroc-nemius muscle with 0.1-ms pulses of current using a stimulus

�Fig. 4 ZIKV can broadly infect the brain of adult, AG129 mice. (A)Sagittal section of the brain of the animal having the most pervasiveCNS ZIKV infection, labeled with antibodies against GFAP (magenta)and ZIKV (green). (B, C) Examples of ZIKV infection (green) in thehippocampus of sham (B) and ZIKV (C, close-up of box in A) infectedanimals. Outlined region in B shows the region that was quantified in N.(D–K) Examples of sham (D, F, H, J) and ZIKV (E, G, I, K) infectedmotor cortex labeled with antibodies against GFAP (astrocytes, magenta)and ZIKV (green) (D, E), CD3 (T cells, green) (F, G), Ly6G (neutrophils,green) (H, I), and iba1 (microglia/macrophages, green) (J, K). Box in Dshows the location and size of the region that was quantified in L,M, P–S.(E1–E3) Close-up of infected astrocytes in boxed area of E showing theGFAP (E1) and ZIKV (E2) channels separately and merged (E3).Arrowheads mark ZIKV+, GFAP+ astrocytes. (G1) Close-up of T cellsin boxed area of G. (I1) Close-up of neutrophils in boxed area of I. (J1)Close-up of resting microglia in boxed area of J. (L, M, P–S)Quantification of signal in a region of motor cortex containing the layerV pyramidal neurons (corresponding to the boxed region shown in D).Each dot represents one animal. Bars indicate mean and SEM of eachgroup. *p = 0.0315; **p = 0.0088; #p = 0.0221; @p = 0.0102. (N)Quantification of signal in the pyramidal cell layer of the hippocampus(corresponding to the outlined region shown in B). (O) Correlation be-tween ZIKV infection level in the hippocampus pyramidal cell layer andseverity of seizure-like activity. Line represents linear regression analysis.Scale bars = 1 mm (A), 500 μm (B, C same scale), 250 μm (D–K samescale), 50 μm (E1–E, G1, I1, J1 same scale)

278 J. Neurovirol. (2018) 24:273–290

0.0

0.2

0.4

0.6

seizure severity

hip

po

cam

pu

s p

yram

idal

ce

ll in

fect

ion

leve

l(z

ika+

are

a/ p

yram

idal

cel

l are

a)

none

mild

seve

re

0.00

0.02

0.04

0.06

0.08

0.10

cort

ex in

fect

ion

leve

l(Z

IKV

+are

a/ w

hole

are

a)

0.0

0.2

0.4

0.6

astr

ocy

te in

fect

ion

leve

l(Z

IKV

+GF

AP

+are

a/ G

FA

P+

area

)

0.000

0.005

0.010

0.015

0.020

0.025

astr

og

liosi

s(G

FA

P+

area

/ who

le a

rea)

0.00

0.01

0.02

0.03

neu

tro

ph

il in

vasi

on

(Ly6

G+

area

/ who

le a

rea)

0.000

0.005

0.010

0.015

0.020

T-c

ell i

nva

sio

n(C

D3+

are

a/ w

hole

are

a)

0.000

0.005

0.010

0.015

0.020

0.025

mic

rog

lia/

mac

rop

hag

e ac

tivat

ion

(iba1

+ ar

ea/ w

hole

are

a)

0.0

0.2

0.4

0.6

hip

po

cam

pu

s p

yram

idal

ce

ll in

fect

ion

leve

l(Z

IKV

+are

a/ p

yram

idal

cel

l are

a)

sham

ZIK

Vsham

ZIKV

ba

c

d

e

f

g

h

i

j

k

e1 e2 e3 g1 l1 j1

GFAP ZIKV

CD3 nuclei Ly6G nuclei

GFAP ZIKV nuclei

GFAP ZIKV iba1 nuclei

l m n o

p q r s

ZIKVsham

GFAP ZIKV merge

**

#*

*

@

J. Neurovirol. (2018) 24:273–290 279

isolator (WPI Isostim A320). For lumbosacral spinal cordstimulation, monopolar needle electrodes (BIOPACSystems, EL452) were used, and the anode was inserted be-tween the T13 and L1 vertebrae, and the cathode was insertedbetween the L1 and L2 vertebrae (Basoglu et al. 2013;Harrison et al. 2013). No laminectomy was necessary.Muscle responses were recorded with 4-mm-diametershielded Ag/AgCl surface recording electrodes (EL254S,Biopac system) filled with electrode gel. Electrodes wereplaced on the skin at the belly of the muscle and at the ante-rior aspect of the ankle and connected to a differential ampli-fier (DAM 50, WPI) with a gain of ×100 and filtered with a300-1K Hz band pass filter. Data were acquired at a 20 K/s

sampling rate with Powerlab 4/25 and LabChart 8 software(ADInstruments). Responses to five pulses at 1 Hz were av-eraged, and the current was increased incrementally until amaximum amplitude was reached. Maximal CMAP ampli-tudes were measured from peak to peak of the M wave.

For experiments with α-btx, uninfected, male AG129mice, about 2.5 months of age, were used. After finding thestimulation current that gave the maximal CMAP response,50 μl of 12.5, 6.25, or 3.13 μM α-btx (Biotium, Inc., #00010-1) was injected into the gastrocnemius muscle (with the gelpads kept in place) with a 26G needle (Nakanishi et al. 2005).CMAP responses after maximal stimulation current were re-corded at intervals for up to 1 h afterα-btx injection. Dilutions

0

10

20

30

40

50

MN

su

rviv

al(#

neu

rons

in v

entra

l hor

ns)

L4-5 S0.0

0.5

1.0

1.5

MN

su

rviv

al(#

MN

/ave

in s

ham

s at

eac

h sp

inal

leve

l)

**

**

**

0.00

0.05

0.10

0.15

0.20

0.25

astr

og

liosi

s(G

FA

P+

area

/ who

le a

rea)

0.00

0.02

0.04

0.06

0.08

0.10

ZIK

V in

fect

ion

leve

l(Z

IKV

+ ar

ea/ w

hole

are

a)

0.0

0.2

0.4

0.6

0.8

1.0

MN

infe

ctio

n le

vel

(ZIK

V+

neur

ons/

vent

ral h

orn

neur

ons)

0.00

0.05

0.10

0.15

0.20

0.25as

tro

ctye

infe

ctio

n le

vel

(ZIK

V+G

FA

P+

area

/ G

FA

P+

area

)

0.00

0.01

0.02

0.03

0.04m

icro

glia

/m

acro

ph

age

activ

atio

n(ib

a1+

area

/ who

le a

rea)

0.000

0.005

0.010

0.015

0.020

neu

tro

ph

il in

vasi

on

(Ly6

G+

area

/ who

le a

rea)

0

50

100

150

200

250

T-c

ell i

nva

sio

n(#

CD

3+ c

ells

/mm

2) ZIKVsham

a

b

b1 b2 b3 d1 d2 d3 f1 h1 i1

c

d

e

f

g i

j

ChAT ZIKV

sham

ZIK

V

GFAP ZIKV CD3 nuclei Ly6G nuclei iba1

ChAT ZIKV merge GFAP ZIKV merge

p

h

** ## *

****###

***

k l m n o

q r s

Fig. 5 ZIKV heavily and consistently infects the lumbosacral spinal cordof adult, AG129mice. (A–J) Examples of sham (A, C, E, D, I) and ZIKV(B, D, F, H, J) infected lumbar spinal cords (L4-5 level) labeled withantibodies against ChAT (motor neurons, magenta) and ZIKV (green)(A, B), GFAP (astrocytes, magenta) and ZIKV (green) (C, D), CD3 (Tcells, green) (E, F), Ly6G (neutrophils, green) (G, H), and iba1(microglia/macrophages, white) (I, J). (B1–B3) Close-up of infected mo-tor neurons in boxed area of B showing the ChAT (B1) and ZIKV (B2)channels separately and merged (B3). Arrowheads mark ZIKV−, ChAT+neurons; arrows mark ZIKV+, ChAT+ neurons, and asterisks mark

ZIKV+ neurons with low levels of ChAT. (D1–D3) Close-up of infectedastrocytes in boxed area of D showing the GFAP (D1) and ZIKV (D2)channels separately and merged (D3). Arrowheads mark ZIKV+, GFAP+astrocytes. (F1) Close-up of T cells in boxed area of F. (H1) Close-up ofneutrophils in boxed area of H. (I1) Close-up of restingmicroglia in boxedarea of I. (K–S) Quantification of images. Each dot represents one animal.Bars indicate mean and SEM of each group. **p = 0.0088; ##p = 0.0014;*p = 0.0104; ****p < 0.0001; ***p = 0.0004; ###p = 0010. Scalebars = 500 μm (A–J same scale), 100 μm (B1–B3 same scale), 50 μm(D1–D3 and F1–I1 same scale)

280 J. Neurovirol. (2018) 24:273–290

of α-btx were made with saline, and thus, a 50 μl saline in-jection was used in the control.

Statistics

Data were graphed and analyzed with Prism (GraphPadSoftware, Inc.) for statistical significance using t tests ortwo-way ANOVAs with post hoc t tests. Linear regressionwas used for correlation analyses.

Results

Experiment no. 1 disease and behavioral motor deficits

ZIKV-infected animals began losing weight by 7 DPI andstarted succumbing to the disease by 11 DPI (Fig. 1). By 16DPI, all ZIKV-infected animals had died or been humanelyeuthanized.

The onset of hindlimb motor deficits in ZIKV-infectedmice ranged from 8 to 13 DPI (Fig. 2a, b). Initial symp-toms included tail weakness (tail not in the up positionduring walking) and subtle hindlimb lateral skidding andweakness (as indicated by a limp or wobble). Diseaseprogressed quickly after the onset of motor deficits andmorbidity occurred within 2–3 days. The severity of motordeficits generally increased rapidly, with VPS scoresreaching as high as 5 or 6 before death (hindlimb draggingwith little to no joint movement). Mice were grouped intoearly (8–11 DPI) and late (12–13 DPI) onset groups in Fig.2c, d so that scores from later onset mice would not maskthe progression of disease in the early onset mice. Deficitswere more severe on the right side compared with the left,

corresponding to the side of virus inoculation. Forelimbswere rarely affected.

Seizure-like activity was also observed in ZIKV-infectedmice (Fig. 3). Six of 10 ZIKV-infected mice had some seizure-like activity, and 4 mice were scored as severe (Fig. 3a).Seizure-like activity did not correlate with VPS score (Fig.3b). Half of the mice in which severe seizure-like activitywas noted had no to only subtle hindlimb deficits after seizureactivity had subsided. Additionally, there were mice withcomplete paralysis (VPS score of 6) that were never observedto have seizure-like activity. Two other disease phenotypes,walking in circles and extreme mobile activity (not shown),were also observed (data not shown).

Interestingly, the severity of hindlimb paralysis andseizure-like activity did not correlate well with changes inbody weight (Supplementary Fig. S1), suggesting that mech-anisms affecting neuroinvasion might operate independentlyfrom those affecting systemic infection and body weightchanges.

Experiment no. 1 immunohistofluorescence

Initial histopathologic analysis of our relatively thick, frozensections with hematoxylin and eosin staining was consideredto be preliminary. Nevertheless, the following preliminary re-sults were evidence of mild to moderate, multifocal encepha-litis in the brain; moderate to severe, multifocal myelitis in thespinal cord; and no lesions in the gastrocnemius muscle orsciatic nerve (data not shown). To further identify anatomicalfeatures that might be associated with motor deficits, we per-formed immunohistofluorescent analyses of the motor cortex,lumbosacral spinal cord, sciatic nerve, and gastrocnemius

0.00

0.05

0.10

0.15

0.20

ZIK

V in

fect

ion

leve

l(z

ika+

are

a/ w

hole

are

a)

0

100

200

300

# un

mye

linat

ed a

xons

0

500

1000

1500

2000

2500

# m

yelin

ated

axo

ns

0

10

20

30

40

# em

pty

myl

ein

shea

ths

a c

b d

e f

g h

ZIKVsham

MBPNF

ZIKV

sham

ZIK

V

Fig. 6 ZIKV can infect sciatic nerves of adult, AG129 mice, but axonsand myelin appear healthy. a–d Examples of sham (a, c) and ZIKV (b, d)infected sciatic nerves labeled with antibodies against MBP (myelin,magenta) and NF (axons, green) (a, b) and ZIKV (white) (c, d). Grid ina shows the counting grid used for quantification in f–h. e–h

Quantification of ZIKV signal (e) and number of myelinated axons (f),unmyelinated axons (g), and empty myelin sheaths (h) in the whole crosssection of nerve. Each dot represents one animal. Bars indicate mean andSEM of each group. Scale bar = 50 μm (a–d same scale)

J. Neurovirol. (2018) 24:273–290 281

muscle. The kidney, liver, and pancreas were collected to as-sess systemic infection level.

Motor cortex

Using a polyclonal antibody against ZIKVenvelope glyco-protein, ZIKV immunoreactive (ir) was seen in all parts ofthe adult AG129 mouse brain, including motor cortex, cer-ebellum, hippocampus, and brainstem (Fig. 4 (A)). In themotor cortex, ZIKV infects both neurons and astrocytes(arrows in Fig. 4 (E1–E3)), and many of the neurons havethe morphology and location of layer V pyramidal neuronswhich are upper motor neurons that project to the spinalcord (Fig. 4 (E and E1–E3)). Quantification of a region ofthe motor cortex containing layer V (shown in box in Fig. 4(D)) revealed that 1–2% and up to 6–8% of the area wereZIKV ir in ZIKV-infected mice (Fig. 4 (L)). Additionally,20–50% of astrocytes in this area were ZIKV ir (Fig. 4(M)), as measured by pixel-based quantification.

There were also significant increases in astrogliosis (Fig. 4(E, E1–E3, and P)), T cell infiltration (Fig. 4 (G, G1, and Q)),and neutrophil infiltration (Fig. 4 (I, I1, and R)) in the ZIKV-infectedmotor cortex compared to sham infected (Fig. 4 (D, F,and H)). Interestingly, microglia and macrophages were virtu-ally undetectable in the ZIKV-infected AG129 mouse brain(Fig. 4 (K and S)), though resting microglia were apparent insham-infected brains (Fig. 4 (J and J1).

Hippocampus

The pyramidal cell layer of the hippocampus was also strik-ingly ZIKV ir in many ZIKV-infected mice (Fig. 4 (C)).Quantification of this region (outlined in Fig. 4 (B)) demon-strated that 20–50% of this layer was ZIKV ir in three of eightmice (Fig. 4 (N)). Because of the association between thehippocampus and seizure activity, the severity of seizure-likeactivity was plotted against hippocampus pyramidal cell layerZIKV infection levels, but no correlation was found (Fig. 4(O)). Likewise, no correlation was found between motor cor-tex infection level and seizure-like activity (data not shown).

Lumbosacral spinal cord

ZIKV ir was also abundant in the lumbosacral spinal cordwhich contains lower motor neurons that innervate thehindlimbs (Fig. 5 (A–D)). Overall, 1–7% of the cross-sectional area was ZIKV ir (Fig. 5 (K)). As with the brain,astrocytes (arrowheads in Fig. 5 (D1–D3 and M)) and neuronswere infected. Using a ChAT antibody to label ventral motorneurons, many were found to be ZIKV ir (arrows in Fig. 5(B1–B3)) and a few ventral cells with clear neuronal morphol-ogy were strongly ZIKVir, but only weakly ChAT ir (asterisksin Fig. 5 (B1–B3)). As many as 20–70% of ventral motor

neurons were ZIKV ir (Fig. 5 (L)), which suggests thatZIKV can infect and replicate in lower motor neurons in thespinal cord. The number of motor neurons at the L4-L5 and Slevels was also diminished in many ZIKV-infected mice (Fig.5 (N and O)), but the difference from sham-infected mice didnot reach statistical significance, likely because the diseaseprogressed so rapidly.

As in the brain, astrogliosis (Fig. 5 (C, D, and P)), T cellinfiltration (Fig. 5 (E, F, F1, and Q), and neutrophil infiltration(Fig. 5 (G, H, H1, and R) were markedly increased in ZIKV-infected mice compared to sham-injected mice. Conversely,microglia and macrophages were markedly reduced or absent(Fig. 5 (J and S)) in these ZIKV-infected AG129 mice, asobserved in the brain, though resting microglia were evidentin sham-infected spinal cords (Fig. 5 (I and I1).

Sciatic nerve

ZIKV ir reactivity was detected in cross sections of the sciaticnerve (Fig. 6c–e); however, this did not appear to have astatistically significant effect on the number of myelinatedaxons (Fig. 6a, b, f), unmyelinated axons (Fig. 6g), or emptymyelin sheaths (Fig. 6h).

Gastrocnemius muscle, kidney, liver, and pancreas

ZIKV ir and neutrophil infiltration were not detected in thegastrocnemius muscles of ZIKV-infected mice (data notshown). Infection of lower motor neurons in the spinal cordand their axons in the sciatic nerve did not appear to lead to astatistically significant loss of NMJ number or innervation(Fig. 7). No ZIKV ir was detected in the kidney, liver, orpancreas of ZIKV-infected mice (data not shown).

Correlations

To determine which anatomical features might be correlatedwith hindlimb motor deficits, we used our data to make pre-liminary correlations between histological parameters of thespinal cord (Fig. 8a–f), motor cortex (Fig. 8g–k), NMJ (Fig.8l), and sciatic nerve (Fig. 8m) and VPS scores. Statisticallysignificant correlations were only found with spinal cord pa-rameters. Increased numbers of ZIKV-infected ventral motorneurons in the lumbosacral spinal cord are associated withhigher VPS scores (Fig. 8a, p < 0.05), and higher VPS scoresare associated with decreased survival of spinal motor neurons(Fig. 8c, p < 0.001). These data suggest that ZIKV infection oflower motor neurons and lower motor death adversely affectshindlimb function. Interestingly, the level of neutrophil infil-tration was inversely proportional to VPS score (Fig. 8f,p < 0.05), which suggests that the presence of neutrophilsmight mitigate motor deficits. To get a preliminary indicationof how neutrophil invasion and spinal motor neuron viral

282 J. Neurovirol. (2018) 24:273–290

0 1 2 3 4 5 60.0

0.2

0.4

0.6

0.8

VPS score

spin

al m

oto

r n

euro

n in

fect

ion

leve

l

*

0 1 2 3 4 5 6

0.10

0.15

0.20

0.25

VPS score

spin

al c

ord

astr

og

liosi

s

0 1 2 3 4 5 60.000.050.100.150.200.25

VPS score

spin

al c

ord

ast

rocy

te

infe

ctio

n le

vel

0 1 2 3 4 5 60.000

0.005

0.010

0.015

0.020

VPS scoresp

inal

co

rd

neu

tro

ph

il in

vasi

on

#

0 1 2 3 4 5 60.0

0.5

1.0

1.5

VPS score

spin

al m

oto

r n

euro

n

surv

ival **

0 1 2 3 4 5 60

50100150200250

VPS score

spin

al c

ord

T-

cell

inva

sio

n

0 1 2 3 4 5 60.000.020.040.060.080.10

VPS score

mo

tor

cort

exin

fect

ion

leve

l

0 1 2 3 4 5 60.0

0.2

0.4

0.6

VPS score

mo

tor

cort

ex a

stro

cyte

in

fect

ion

leve

l

0 1 2 3 4 5 60.0000.0050.0100.0150.0200.025

VPS score

mo

tor

cort

exas

tro

glio

sis

0 1 2 3 4 5 60.00

0.01

0.02

0.03

VPS score

mo

tor

cort

ex

neu

tro

ph

il in

vasi

on

0 1 2 3 4 5 60.000

0.005

0.010

0.015

0.020

VPS score

mo

tor

cort

ex

T-ce

ll in

vasi

on

0 1 2 3 4 5 60.00.20.40.60.81.0

VPS score

NM

J in

ner

vatio

n

0 1 2 3 4 5 60.00

0.05

0.10

0.15

0.20

VPS score

ner

ve in

fect

ion

leve

l

k

ZIKV malesZIKV females

a b c

d e f

g h

i j

l m

Fig. 8 Motor deficits correlatewith spinal motor neuroninfection level, spinal motorneuron survival, and spinal cordneutrophil infiltration. a–m Imagequantification parameters shownin Figs. 4, 5, 6, and 7 graphedagainst the highest VPS scoreseen in the animal. Each dotrepresents one ZIKV-infectedanimal (gray = female,black = male). Lines representlinear regression analysis.*p = 0.0317; **p = 0.0071;#p = 0.0164

0.0

0.2

0.4

0.6

0.8

1.0

NM

J in

ner

vatio

n(N

F+

NM

J/ to

tal N

MJ)

0

50

100

150

# N

MJ

(tot

al #

cou

nted

)

ZIKVsham

b

c

dsh

amZ

IKV

btx NF btx NF

a a1 a2

b1 b2

Fig. 7 NMJs appear normal after ZIKV infection of adult, AG129 mice.(A, B) Examples of sham (A) and ZIKV (B) infected gastrocnemiusmuscle labeled with TMR-α-btx (NMJ endplate, magenta) and an anti-body against NF (axons, green). (A1, B1) α-btx channel alone. (A2, B2)NF channel alone. Arrowheads mark innervated NMJs (α-btx+ and NF+

). (C, D) Quantification of percent of NMJs that are innervated (C) andtotal number of NMJs counted (D). Each dot represents one animal. Barsindicate mean and SEM of each group. Scale bar = 50 μm (A, B samescale)

J. Neurovirol. (2018) 24:273–290 283

infection are related to spinal motor neuron survival (as op-posed to VPS score), correlation graphs with these parameterswere prepared as well and similar trends were observed(Supplemental Fig. S2). VPS scores did not appear to correlatewith virus levels in the motor cortex (Fig. 8g, h) or with sciaticnerve infection levels (data not shown).

Overall, these results suggest that hindlimb motor deficitsin these AG129mice during the acute phase of ZIKVinfectionare likely caused primarily by spinal cord myelitis, thoughupper motor neuron disease and peripheral neuropathy maycontribute.

Experiment no. 2 disease and behavioral motor deficits

Experiment no. 2 reproduced the disease and motor pheno-types seen in experiment no. 1. ZIKV-infected animals beganlosing weight by 7 DPI and started succumbing to the diseaseby 11 DPI (Supplemental Fig. S3). All ZIKV-infected micehad died or been humanely euthanized by 14 instead of 16DPI. VPS assessments were done less frequently, but symp-toms appeared at a similar timing and progressed at a similarrate (Supplemental Fig. S4).

Experiment no. 2 electrophysiological assessment

When mice in experiment no. 2 had VPS scores ≥ 3, gastroc-nemius muscle CMAPs were recorded in response to stimu-lation at the sciatic notch, followed by stimulation of the lum-bosacral spinal cord to further delineate the cause of thehindlimbmotor deficits. We reasoned that since motor deficitsprogress quickly after symptom onset and many motor neu-rons, though infected, have not died by the time severe deficitsoccur, axons and NMJs may still be healthy even if the neu-ronal cell body is sick. Thus, CMAP amplitudes may be nor-mal in response to sciatic notch stimulation, but decreasedupon lumbosacral spinal cord stimulation. Responses to stim-ulation at these two sites, therefore, might distinguish betweenmyositis/neuritis and myelitis.

Because CMAP measurement in response to lumbar spinalcord stimulation was an invasive procedure, pre-infectionbaselines were only obtained in response to sciatic notch stim-ulation. Figure 9a shows the CMAP amplitudes for each ani-mal pre-infection and post-infection in response to stimulationat the two sites. There are no statistically significant differ-ences between groups when these Babsolute^ values are used.The use of relative values, however, are more compelling, andmay be necessary because individual differences in the thick-ness and composition of tissue between the surface of the skinand the muscle can affect CMAP amplitudes (Nordander et al.2003). When post-infection CMAP amplitude was expressedrelative to pre-infection amplitude (both in response to sciaticnotch stimulation), there was no statistically significant differ-ence between the ZIKV-infected and sham-infected groups

(Fig. 9b, Bpre minus post, at notch^), suggesting that the axonsand NMJs are healthy after infection and motor deficits are notdue to myositis or neuritis. When post-infection CMAP am-plitude in response to stimulation of the spinal cord wasexpressed relative to the amplitude in response to stimulationat the sciatic notch, however, there was a statistically signifi-cant relative decrease in amplitude in the ZIKV-infected group(Fig. 9b, Bpost, at s. cord minus notch,^ p < 0.05). This sug-gests that, after infection and at the time motor deficits areobserved, the axons and NMJs are healthy enough to generatea normal CMAP response when axons are stimulated at thesciatic notch, but the neuronal cell bodies in the spinal cord areimpaired and less able to generate normal CMAP responseswhen stimulated at the lumbosacral spinal cord. This indicatesthat myelitis, rather thanmyositis or neuritis, contributes to themotor deficits. The fact that VPS score did not correlate withthe relative CMAP amplitude in response to spinal cord stim-ulation (Fig. 9c), however, suggests that other factors alsocontribute to hindlimb motor deficits.

To validate that the CMAP indeed reflected muscle activ-ity, and not just nerve activity, an inhibitor of the NMJ nico-tinic acetylcholine receptor, α-btx, was injected into the gas-trocnemius muscles of uninfected adult AG129 mice, andCMAPs were recorded for up to 1 h after injection. CMAPamplitudes decreased in a dose-dependent manner over thecourse of the 1 h after α-btx injection, whereas CMAP ampli-tude in the saline-injected animal was relatively stable(Fig. 10). These data suggest that the CMAP reflects mostlymuscle rather than nerve activity, and thus, CMAP measuredthe health status of all structures between the point of stimu-lation and the muscle (including the nerve and NMJ).

Discussion

This study demonstrates that a contemporary strain of ZIKV(PRVABC59) can widely infect astrocytes and neurons in thebrain and spinal cord of adult AG129 mice and cause rapidlyprogressing hindlimb paralysis, as well as severe seizure ac-tivity, during the acute phase of disease. Motor deficits inthese circumstances appear to be primarily due to myelitisand possibly encephalitis as opposed to a peripheral neuropa-thy or GBS-like syndrome. This is an important new findingbecause the most severe histopathologies previously reportedin AG129 mice were in the brain and muscle (Aliota et al.2016). The finding that myelitis is likely the primary cause ofZIKV-induced paralysis is also in alignment with what hasbeen seen with other flavivirus-induced motor deficits(Sejvar et al. 2003; Solomon et al. 1998). This conclusion isalso supported by our data suggesting that muscle and periph-eral nerve functions are normal. Electrophysiological CMAPamplitude measurements of sciatic nerve and sural muscleswere normal in response to stimulation at the sciatic notch.

284 J. Neurovirol. (2018) 24:273–290

Additionally, the anatomy of the sciatic nerve and gastrocne-mius muscle, as revealed by neurofilament/MBP andneurofilament/NMJ staining, respectively, was minimallydisrupted. In contrast, ZIKV ir was high, intense, and obviousin the brain and spinal cord. The severity of hindlimb motordeficits correlated with increased numbers of ZIKV-infectedlumbosacral spinal motor neurons and decreased numbers ofspinal motor neurons. Relative CMAP amplitudes upon

stimulation at the lumbar spinal cord were also reduced whenobvious motor deficits were present (VPS score ≥ 3). ZIKVinfections in the brain and spinal cord were also associatedwith astrogliosis as well as T cell and neutrophil infiltration.Thus, acute ZIKV infection of adult AG129 mice may be auseful model for studying the mechanisms of ZIKV-inducedmyelitis (Anaya et al. 2017; Dirlikov et al. 2016), encephalitis,and seizure activity.

Myelitis

The results of this study may be clinically relevant, becausethere have been several reports of myelitis during recent ZIKVoutbreaks. In the February 2016 outbreak in Guadeloupe, a15-year-old girl developed hemiparalysis that was associatedwith detection of ZIKV in the cerebrospinal fluid. Magneticresonance imaging revealed extensive lesions in the cervicaland thoracic spinal cord (Mecharles et al. 2016). Investigatorsreporting on ZIKV-associated GBS in Puerto Rico noted 26cases of neurologic disorders other than GBS, 2 of which were

0

20

40

60

CM

AP

am

plitu

de (

mV

)

1F-251F-2001F-2511F-252

1M-891M-901M-773

1M-192

1M-24

1M-196

1M-197

2F-1982F-253

2M-922M-772

2M-1932M-194

shamfemale

ZIKVmale

ZIKVfemale

shammale

a b

-40

-20

0

20

CM

AP

am

plitu

de (

mV

)

pre

min

us p

ost,

at n

otch

post

, at s

.cor

d

min

us n

otch

*

0 1 2 3 4 5 6-40-30-20-10

010

VPS score(on side measured)

CM

AP

am

p (m

V)

(pos

t, at

s.c

ord

min

us n

otch

)

c

Fig. 9 Relative gastrocnemius muscle CMAP amplitudes in response tolumbosacral spinal cord stimulation are reduced after ZIKV infection. aGastrocnemius muscle CMAP amplitude pre-infection in response tosciatic notch stimulation (pre, notch); post-infection in response to sciaticnotch stimulation (post, notch); post-infection in response to lumbosacralspinal cord stimulation (post, s. cord). Post-infection measurements wereobtained when VPS score ≥ 3. b Difference between CMAP amplitudes(also referred to as Brelative^ CMAP amplitudes). For Bpre minus post, atnotch,^ the pre-infection value was subtracted from the post-infection

value in response to sciatic notch stimulation. For Bpost, s. cord minusnotch,^ the post-infection value in response to sciatic notch stimulationwas subtracted from the value in response to lumbosacral spinal cordstimulation. *p = 0.0355. c Relative CMAP amplitudes in response tolumbosacral spinal cord stimulation (post, s. cord minus notch in b) plot-ted against the VPS score on the measured side. Each symbol representsone animal. Bars indicate mean and SEM of each group. Lines in a and bconnect measurements from same animal (symbols of individual animalsare connected). Line in c represents linear regression analysis

0 20 40 600

20

40

60

min after -btx injection

CM

AP

am

plitu

de

(mV

)

-btx, 12.5 µMsaline

-btx, 6.25 µM-btx, 3.13 µM

0 20 40 600

20406080

100

min after -btx injection

CM

AP

am

plitu

de

(% b

asel

ine)

BA

Fig. 10 Gastrocnemius muscle activity contributes to CMAPs. aGastrocnemius CMAP amplitudes from uninfected AG129 male micebefore and after injections of α-btx. b CMAP amplitudes expressed as apercentage of baseline

J. Neurovirol. (2018) 24:273–290 285

myelitis (Dirlikov et al. 2016). In a case-controlled study inColombia from a national population-based surveillance sys-tem (Anaya et al. 2017), 29 patients with ZIKV-associatedGBS were compared to matched controls. Thirteen of thesepatients were reported to have other neurological syndromes:three with myelitis, three with peripheral facial palsy, six withtransverse myelitis, and one with thoraco-lumbosacralmyelopathy.

The observation that ZIKV can infect and kill spinalmotor neurons and that this correlates with hindlimb motordeficits is a unique finding of this report and may be clin-ically important. Pathology reports of human ZIKV infec-tion are limited, but do reveal neuronophagia in the brain(Solomon et al. 2016), which suggests that ZIKV may becapable of eliciting neuron death in human neurologicaltissues as well as in the AG129 mouse. ZIKV infectionhas previously been reported in the spinal cords ofAG129 mice (Julander et al. 2017; Zmurko et al. 2016),and ZIKV-induced histopathology or myelitis has been ob-served in IFNAR−/− (Lazear et al. 2016) and Swiss Webstermice infected as newborns (Fernandes et al. 2017).

Myelitis in other flavivirus infections

The ZIKV-induced motor deficits in AG129mice are reminis-cent of motor deficits caused by other flaviviruses, such asWNV, in immunocompetent mice and hamsters (Morreyet al. 2004; Morrey et al. 2008; Shrestha et al. 2003;Siddharthan et al. 2009; Xiao et al. 2001; Zukor et al. 2017).These flavivirus models also demonstrate virus infection andkilling of spinal motor neurons together with mild to severesigns of motor impairment (Zukor et al. 2017).

Encephalitis

The vast majority of adults infected with ZIKV do not developsevere neurological disease such as encephalitis, yet theAG129 mice of this study are immunodeficient and developencephalitis andmyelitis. As such, the AG129mouse model isprobably more relevant to infection of immunocompromisedpatients. Some clinical reports of infected immunocompro-mised patients describe severe neurological complications(Henry et al. 2017; Schwartzmann et al. 2017). For example,an immunosuppressed heart transplant patient developedacute neurological impairment and mental deterioration afterZIKV infection, which culminated in death. Autopsy revealedZIKV infection of the brain and cerebral spinal fluid by RT-PCR, immunohistochemistry, and electron microscopy. Thehistopathology of the brain was consistent with meningoen-cephalitis (Schwartzmann et al. 2017).

The fact that relative CMAP amplitude after spinal cordstimulation did not statistically correlate with VPS scoresuggests that damage upstream of spinal motor neurons

may contribute to the hindlimb deficits. For example,CMAP amplitude was not reduced in two infected micewith overt paralysis (VPS score 5–6, Fig. 9c). Paralysisin these animals could be caused by motor cortex, cerebel-lum, or brainstem dysfunction, as extensive infection wasobserved in these areas of the brain. While upper motorneuron disease is associated with rigid, rather than flaccidparalysis, we occasionally saw symptoms that could beinterpreted as rigidity, such as walking with high haunchesand extended/stiff limbs, but it was difficult to reliablyassess this with the VPS test. Another possible explanationfor the paralysis in animals with normal relative CMAPamplitudes is that spinal motor neuron infection damagesdendrites and post-synaptic densities such that the neuronsare not able to receive signals physiologically, but whenstimulated externally with an electrode, the axon hillockand axon are healthy enough to propagate an action poten-tial to the muscle.

Seizures

The observation of seizure-like activity in infected AG129mice in this study and in immunocompetent mice infected asneonates in another study (Manangeeswaran et al. 2016) mayhave clinical relevance since recent ZIKV infections havebeen associated with seizures in humans (Asadi-Pooya 2016;Pastula et al. 2016). Notably, the seizure-like activity in thisstudy was not correlated with the VPS score, which suggeststhat the seizure activity was not simply a manifestation ofmotor deficits. To further investigate the relevance, seizureswould need to be confirmed in ZIKV-infected mice with moredetailed analyses.

Peripheral neuropathy

Motor deficits seen in this study are not likely to be theresult of peripheral neuropathy. While ZIKV ir was ob-served in the sciatic nerve, axon and myelin morphologydid not appear to be altered. ZIKV also did not appear toalter the function of the sciatic nerve, which is supportedby the observation that gastrocnemius CMAP amplitudesin response to sciatic notch stimulation were not affected.The observation of viral antigen in axons is consistent withprior findings that four families of viruses, includingflaviviruses, can undergo axonal transport and spread inthe nervous system (Samuel et al. 2007; Taylor andEnquist 2015; Wang et al. 2009), as well as the finding thatZIKV appears to be able to travel in axons (van den Polet al. 2017). We were not able to determine what effectZIKV might have on the development of peripheral neu-ropathies, such as GBS, in this study because the mice dieduring the acute infection and peripheral neuropathies tendto develop in chronic stages.

286 J. Neurovirol. (2018) 24:273–290

Cellular immune responses

In the current study, inflammatory cellular responses, such asastrogliosis, neutrophil infiltration, and Tcell infiltration, werepresent in the brain and spinal cord. The possible influence ofeach of these responses to disease severity, whether causativeor protective, is explored below. Future work should compre-hensively establish correlation with larger sample sizes andthen determine which relationships are causative or protective,so that the cellular mechanisms underlying ZIKV-inducedmotor deficits can be understood.

Neutrophils

Results from this study suggest that neutrophils play a role inmitigating or protecting ZIKV-infected AG129 mice frommotor deficits, probably by enhancing immune responses thatreduce viral load. Increased numbers of neutrophils, as detect-ed by the Ly6G antibody, were inversely correlated with VPSscore (p < 0.05). Studies with WNV demonstrate that neutro-phils can play complex roles during infection. They appear tobe protective early in WNV infection, but detrimental later.When neutrophils were depleted before WNV challenge inC3H/HeJ mice, mortality increased, but when they were de-pleted after WNV challenge, mortality decreased (Bai et al.2010). Another level of complexity regarding the role of neu-trophils involves mosquito bites.When neutrophils are deplet-ed and inflammasome activity is blocked, the ability of mos-quito bites to promote infection and inflammation is abrogated(Pingen et al. 2016). Therefore, the role of neutrophils inZIKV-induced disease should be further investigated.

Microglia and macrophages

The drastic reduction in iba1 ir (a marker for microglia andmacrophages) in the brains and spinal cords of AG129 miceinfected with ZIKV was a unique finding. It could mean thatmicroglia and macrophages are present, but the iba1 marker isstrongly downregulated, or it could mean that microglia arekilled and macrophage infiltration is inhibited. After WNVinfection in C57BL6 mice, microglia and macrophages arestrongly activated (Zukor et al. 2017). They are also stronglyactivated in wild-type C57BL6 or 129S1/SvImJ mice infectedwith ZIKV intracerebrally at E14.5 (Shao et al. 2016). Thus,the decrease in iba1 ir after ZIKV infection in our study mightbe unique to the AG129 mouse strain and could indicate a rolefor microglia and macrophages in protecting against lethalZIKV infection. For example, M1 macrophages elicit pro-inflammatory responses and enable phagocytosis and killingof pathogens, and M2macrophages function in protective andhomeostasis of the CNS (Kabba et al. 2018). A certain balanceof these functions could be overall protective. To further elu-cidate immunological cellular responses, decreased iba1 ir

cells and modulation of other immune cells should be validat-ed with flow cytometric analysis in future studies.

T cells

T cell infiltration was associated with ZIKV infection in thebrain and spinal cord, though there was no correlation be-tween the level of infiltration and the severity of motor defi-cits. This could indicate that T cells play both beneficial anddetrimental roles (Ousman and Kubes 2012; Prinz and Priller2017). For example, CD8(+) Tcells contribute to resolution ofWNV neurological infection by helping to clear WNV fromneurons (McCandless et al. 2008; Shrestha et al. 2012). T cellsalso play beneficial roles in La Crosse virus infection (Winkleret al. 2017). T cells, however, are the cause of fatal meningo-encephalitis after Tacaribe arenavirus infection (Ireland et al.2017). Future studies will delineate the role of T cells in thedevelopment of motor deficits in ZIKV-infected AG129 mice.

Astrocytes

Astrogliosis in the spinal cord was strongly associated withZIKV infection, and many astrocytes were infected by thevirus. As astrocytes regulate synapse formation, function,and elimination (Chung et al. 2015) and control glutamateexcitotoxicity (Murphy-Royal et al. 2017), this likely had asignificant effect on motor neuron health and function. In amodel of human coronavirus, expression of the primary glu-tamate transporter in the adult CNS, GLT-1, was decreased inastrocytes. This was associated with flaccid hindlimb paraly-sis even in the absence of significant neuronal death. Blockadeof 2-amino-3-(5-methyl-3-oxo-1,2-oxazol-4-yl) propranoicacid (AMPA) receptors with a non-competitive antagonist im-proved hindlimb function (Brison et al. 2011). In mousemodels of experimental autoimmune encephalomyelitis, as-trocytes are implicated in the temporary stripping of excitatorysynapses from motor neurons in the spinal cord, leading tosevere flaccid paralysis in the absence of neuronal cell death(Blakely et al. 2015; Marques et al. 2006). Astrogliosis maynot have correlated with VPS scores in our studies because itmay precede and be required for the onset of motor deficits(Blakely et al. 2015).

Susceptibility to disease

In two independent experiments, female mice were found todie slightly earlier than male mice, though this was not statis-tically significant. Other ZIKV studies with AG129 mice atour institute, however, suggest that males and females areequally susceptible (data not shown). Additionally, experi-ment no. 1 appeared to have two distinct groups of mice withan early versus late onset of symptoms, though this separationwas not distinct in experiment no. 2. We interpret these trends

J. Neurovirol. (2018) 24:273–290 287

(greater female susceptibility and two phases of disease onset)to be the result of stochastics and small sample sizes. Theappearance of two onset groups in experiment no. 1 occurredby chance because we did not sample mice with an interme-diate onset of symptoms.

Conclusion

The unique contributions of this study are that acute ZIKVinfection of adult AG129 mice causes quantifiable, rapidlyprogressing hindlimb motor deficits that often culminates infull paralysis and that these deficits are likely due to myelitisand perhaps encephalitis rather than peripheral neuropathy ormyositis. Extensive histopathology and infection occur in thespinal cord and brain, but not in the sciatic nerve or muscle.The severity of motor deficits is only correlated with infectionand survival levels of lumbosacral spinal motor neurons.Moreover, gastrocnemius CMAP amplitudes upon electricalstimulation of the lumbosacral spinal cord were impaired, butnot upon stimulation of the sciatic nerve. These data shouldguide approaches for understanding ZIKV-induced acute my-elitis or encephalitis in adults.

Acknowledgements We thank the following individuals from theInstitute for Antiviral Research at Utah State University for their excellenttechnical assistance: Mitchell Stevens for help with VPS scoring; BrentHunter for help with immunohistofluorescent staining; Christian Morrillfor help with image quantification; Heidi Julander for help with AG129colony management; Rich Albrechtsen for help with setting up theCMAP assays; Skot Neilson for overseeing animal studies; JaredBennet, Jason Fairbourn, John McClatchy, Devin Pfister, Jean MarieMaxwell, Taylor Redding, Rachelle Stanton, and Trevor Truex forassisting with animal studies; andNeil Motter for maintenance and designof electrophysiological instruments. We also thank Arnaud Van Wettere,DVM, MS, PhD, DACVP at Utah State University’s VeterinaryDiagnostic Laboratory for histopathologic analysis of tissues.

The work was supported by Public Health Service (PHS) researchgrant 1R21NS090133 to J.D.M. from the National Institute ofNeurological Disorders and Stroke (NINDS), National Institutes ofHealth (NIH), and PHS grant 5R33AI101483 and contractHHSN272201000039I/HHSN27200004 to J.D.M. from the VirologyBranch, National Institute of Allergy and Infectious Diseases (NIAID),NIH, and Utah Agriculture Research Station research grant UTA01176 toJ.D.M. The funders had no role in study design, data collection andanalysis, decision to publish, or preparation of the manuscript.

Author contribution KZ, JDM, and JGJ designed the research; KZ,HW, and VS performed the research; KZ, HW, VS, and JDM analyzedthe data; KZ and JDM wrote the paper.

Compliance with ethical standards This work was done in theAssociation for Assessment and Accreditation of Laboratory AnimalCare International-accredited, Biosafety Level 3 Laboratory at UtahState University. All experimental procedures were performed in compli-ance with animal protocols approved by the Institutional Animal Care andUse Committee at Utah State University and in compliance with theBGuide for the Care and Use of Laboratory Animals^ (NationalResearch Council 2011).

Conflict of interest The authors declare that they have no conflict ofinterest.

Open Access This article is distributed under the terms of the CreativeCommons At t r ibut ion 4 .0 In te rna t ional License (h t tp : / /creativecommons.org/licenses/by/4.0/), which permits unrestricted use,distribution, and reproduction in any medium, provided you giveappropriate credit to the original author(s) and the source, provide a linkto the Creative Commons license, and indicate if changes were made.

References

Aguirre S, Maestre AM, Pagni S, Patel JR, Savage T, Gutman D,Maringer K, Bernal-Rubio D, Shabman RS, Simon V, Rodriguez-Madoz JR, Mulder LC, Barber GN, Fernandez-Sesma A (2012)DENV inhibits type I IFN production in infected cells by cleavinghuman STING. PLoS Pathog 8:e1002934

Aliota MT, Caine EA, Walker EC, Larkin KE, Camacho E, Osorio JE(2016) Characterization of lethal zika virus infection in AG129mice. PLoS Negl Trop Dis 10:e0004682

Anaya JM, Rodriguez Y, Monsalve DM, Vega D, Ojeda E, Gonzalez-Bravo D, Rodriguez-Jimenez M, Pinto-Diaz CA, Chaparro P,Gunturiz ML, Ansari AA, Gershwin ME, Molano-Gonzalez N,Ramirez-Santana C, Acosta-Ampudia Y (2017) A comprehensiveanalysis and immunobiology of autoimmune neurological syn-dromes during the Zika virus outbreak in Cucuta, Colombia. JAutoimmun 77:123–138

Asadi-Pooya AA (2016) Zika virus-associated seizures. Seizure 43:13Bai F, Kong KF, Dai J, Qian F, Zhang L, Brown CR, Fikrig E,

Montgomery RR (2010) A paradoxical role for neutrophils in thepathogenesis of West Nile virus. J Infect Dis 202:1804–1812

Basoglu H, Kurtoglu T, Cetin NK, Bilgin MD, Kiylioglu N (2013)Assessment of in vivo spinal cord conduction velocity in rats in anexperimental model of ischemic spinal cord injury. Spinal Cord 51:616–622

Basso DM, Fisher LC, Anderson AJ, Jakeman LB, McTigue DM,Popovich PG (2006) Basso mouse scale for locomotion detects dif-ferences in recovery after spinal cord injury in five common mousestrains. J Neurotrauma 23:635–659

Best SM (2017) The many faces of the flavivirus NS5 protein in antag-onism of type I interferon signaling. J Virol 91

Blakely PK, Hussain S, Carlin LE, Irani DN (2015) Astrocytematricellular proteins that control excitatory synaptogenesis are reg-ulated by inflammatory cytokines and correlate with paralysis sever-ity during experimental autoimmune encephalomyelitis. FrontNeurosci 9:344

Bowen JR, Quicke KM, Maddur MS, O’Neal JT, McDonald CE,Fedorova NB, Puri V, Shabman RS, Pulendran B, Suthar MS(2017) Zika virus antagonizes type I interferon responses duringinfection of human dendritic cells. PLoS Pathog 13:e1006164

Brasil P, Sequeira PC, Freitas AD, Zogbi HE, Calvet GA, de Souza RV,Siqueira AM, de Mendonca MC, Nogueira RM, de Filippis AM,Solomon T (2016) Guillain-Barre syndrome associated with Zikavirus infection. Lancet 387:1482

Brison E, Jacomy H, Desforges M, Talbot PJ (2011) Glutamateexcitotoxicity is involved in the induction of paralysis in mice afterinfection by a human coronavirus with a single point mutation in itsspike protein. J Virol 85:12464–12473

Cao-Lormeau VM, BlakeA,Mons S, Lastere S, Roche C, VanhomwegenJ, Dub T, Baudouin L, Teissier A, Larre P, Vial AL, Decam C,Choumet V, Halstead SK, Willison HJ, Musset L, Manuguerra JC,Despres P, Fournier E, Mallet HP, Musso D, Fontanet A, Neil J,

288 J. Neurovirol. (2018) 24:273–290

Ghawche F (2016) Guillain-Barre syndrome outbreak associatedwith Zika virus infection in French Polynesia: a case-control study.Lancet 387:1531–1539

Cardoso CW, Paploski IA, Kikuti M, Rodrigues MS, Silva MM, CamposGS, Sardi SI, Kitron U, Reis MG, Ribeiro GS (2015) Outbreak ofexanthematous illness associated with Zika, chikungunya, and den-gue viruses, Salvador, Brazil. Emerg Infect Dis 21:2274–2276

Carteaux G, Maquart M, Bedet A, Contou D, Brugieres P, Fourati S,Cleret de Langavant L, de Broucker T, Brun-Buisson C, Leparc-Goffart I, Mekontso Dessap A (2016) Zika virus associated withmeningoencephalitis. N Engl J Med 374:1595–1596

Chung WS, Welsh CA, Barres BA, Stevens B (2015) Do glia drivesynaptic and cognitive impairment in disease? Nat Neurosci 18:1539–1545

Dick GW (1952) Zika virus. II. Pathogenicity and physical properties.Trans R Soc Trop Med Hyg 46:521–534

Dirlikov E, Major CG, Mayshack M, Medina N, Matos D, Ryff KR,Torres-Aponte J, Alkis R, Munoz-Jordan J, Colon-Sanchez C,Salinas JL, Pastula DM, Garcia M, Segarra MO, Malave G,Thomas DL, Rodriguez-Vega GM, Luciano CA, Sejvar J, SharpTM, Rivera-Garcia B (2016) Guillain-Barre syndrome during ongo-ing Zika virus transmission—Puerto Rico, January 1-July 31, 2016.MMWR Morb Mortal Wkly Rep 65:910–914

Fernandes NC, Nogueira JS, Ressio RA, Cirqueira CS, Kimura LM,Fernandes KR, Cunha MS, Souza RP, Guerra JM (2017)Experimental Zika virus infection induces spinal cord injury andencephalitis in newborn Swiss mice. Exp Toxicol Pathol 69:63–71

Harrison M, O’Brien A, Adams L, Cowin G, Ruitenberg MJ, Sengul G,Watson C (2013) Vertebral landmarks for the identification of spinalcord segments in the mouse. NeuroImage 68:22–29

Hatzipetros T, Kidd JD, Moreno AJ, Thompson K, Gill A, Vieira FG(2015) A quick phenotypic neurological scoring system for evalu-ating disease progression in the SOD1–G93Amouse model of ALS.J Vis Exp 104:53257

Henry CR, Al-Attar L, Cruz-Chacon AM, Davis JL (2017) Chorioretinallesions presumed secondary to Zika virus infection in an immuno-compromised adult. JAMA Ophthalmol 135:386–389

Ireland DD, Tami C, Pedras-Vasconcelos J, Verthelyi D (2017) CD4 andCD8 T cells mediate distinct lethal meningoencephalitis in micechallenged with Tacaribe arenavirus. Cell Mol Immunol 14:90–107

Itoh A, HoriuchiM,WakayamaK, Xu J, Bannerman P, Pleasure D, Itoh T(2011) ZPK/DLK, a mitogen-activated protein kinase kinase kinase,is a critical mediator of programmed cell death of motoneurons. JNeurosci 31:7223–7228

Julander JG, Siddharthan V, Evans J, Taylor R, Tolbert K, Apuli C,Stewart J, Collins P, Gebre M, Neilson S, Van Wettere A, LeeYM, Sheridan WP, Morrey JD, Babu YS (2017) Efficacy of thebroad-spectrum antiviral compound BCX4430 against Zika virusin cell culture and in a mouse model. Antivir Res 137:14–22

Kabba JA, Xu Y, Christian H, Ruan W, Chenai K, Xiang Y, Zhang L,Saavedra JM, Pang T (2018) Microglia: housekeeper of the centralnervous system. Cell Mol Neurobiol 38(1):53–71

Lazear HM, Govero J, Smith AM, Platt DJ, Fernandez E, Miner JJ,Diamond MS (2016) A mouse model of Zika virus pathogenesis.Cell Host Microbe 19:720–730

Manangeeswaran M, Ireland DD, Verthelyi D (2016) Zika (PRVABC59)infection is associated with T cell infiltration and neurodegenerationin CNS of immunocompetent neonatal C57Bl/6 mice. PLoS Pathog12:e1006004

Marques KB, Santos LM, Oliveira AL (2006) Spinal motoneuron synap-tic plasticity during the course of an animal model of multiple scle-rosis. Eur J Neurosci 24:3053–3062

McCandless EE, Zhang B, Diamond MS, Klein RS (2008) CXCR4 an-tagonism increases T cell trafficking in the central nervous systemand improves survival from West Nile virus encephalitis. Proc NatlAcad Sci U S A 105:11270–11275

McHanwell S, Biscoe TJ (1981) The localization of motoneurons sup-plying the hindlimb muscles of the mouse. Philos Trans R Soc LondSer B Biol Sci 293:477–508

Mecharles S, Herrmann C, Poullain P, Tran TH, Deschamps N, MathonG, Landais A, Breurec S, Lannuzel A (2016) Acute myelitis due toZika virus infection. Lancet 387:1481

Morrey JD, Day CW, Julander JG, Olsen AL, Sidwell RW, Cheney CD,Blatt LM (2004) Modeling hamsters for evaluating West Nile virustherapies. Antivir Res 63:41–50

Morrey JD, Siddharthan V, Wang H, Hall JO, Skirpstunas RT, Olsen AL,Nordstrom JL, Koenig S, Johnson S, DiamondMS (2008)West Nilevirus-induced acute flaccid paralysis is prevented by monoclonalantibody treatment when administered after infection of spinal cordneurons. J Neurovirol 14:152–163

Murphy-Royal C, Dupuis J, Groc L, Oliet SH (2017) Astroglial glutamatetransporters in the brain: regulating neurotransmitter homeostasisand synaptic transmission. J Neurosci Res 95:2140–2151

Nakanishi ST, Cope TC, Rich MM, Carrasco DI, Pinter MJ (2005)Regulation ofmotoneuron excitability via motor endplate acetylcho-line receptor activation. J Neurosci 25:2226–2232

National Research Council (2011) Guide for the care and use of labora-tory animals. The National Academies Press, Washington, DC

Nicastri E, Castilletti C, Balestra P, Galgani S, Ippolito G (2016) Zikavirus infection in the central nervous system and female genital tract.Emerg Infect Dis 22:2228–2230

Nordander C, Willner J, Hansson GA, Larsson B, Unge J, Granquist L,Skerfving S (2003) Influence of the subcutaneous fat layer, as mea-sured by ultrasound, skinfold calipers and BMI, on the EMG ampli-tude. Eur J Appl Physiol 89:514–519

Ousman SS, Kubes P (2012) Immune surveillance in the central nervoussystem. Nat Neurosci 15:1096–1101

Pastula DM, Durrant JC, Smith DE, Beckham JD, Tyler KL (2016) Zikavirus disease for the neurointensivist. Neurocrit Care 26:457–463

Paxinos G, Franklin KBJ (2004) The mouse brain in stereotaxic coordi-nates: compact, 2nd edn. Academic Press, an imprint of ElsevierLtd, San Diego

Pingen M, Bryden SR, Pondeville E, Schnettler E, Kohl A, Merits A,Fazakerley JK, Graham GJ, McKimmie CS (2016) Host inflamma-tory response to mosquito bites enhances the severity of arbovirusinfection. Immunity 44:1455–1469

Prinz M, Priller J (2017) The role of peripheral immune cells in the CNSin steady state and disease. Nat Neurosci 20:136–144

Rossi SL, Tesh RB, Azar SR, Muruato AE, Hanley KA, Auguste AJ,Langsjoen RM, Paessler S, Vasilakis N, Weaver SC (2016)Characterization of a novel murine model to study Zika virus. AmJ Trop Med Hyg 94:1362–1369

Samarasekera U, Triunfol M (2016) Concern over Zika virus grips theworld. Lancet 387:521–524

Samuel MA, Wang H, Siddharthan V, Morrey JD, Diamond MS (2007)Axonal transport mediates West Nile virus entry into the centralnervous system and induces acute flaccid paralysis. Proc NatlAcad Sci U S A 104:17140–17145

Schneider CA, Rasband WS, Eliceiri KW (2012) NIH image to ImageJ:25 years of image analysis. Nat Methods 9:671–675

Schuler-Faccini L, Ribeiro EM, Feitosa IM, Horovitz DD, Cavalcanti DP,Pessoa A, Doriqui MJ, Neri JI, Neto JM, Wanderley HY, CernachM, El-Husny AS, Pone MV, Serao CL, Sanseverino MT (2016)Possible association between Zika virus infection and microcepha-ly—Brazil, 2015. MMWR Morb Mortal Wkly Rep 65:59–62

Schulz KS, Mossman KL (2016) Viral evasion strategies in type I IFNsignaling—a summary of recent developments. Front Immunol 7:498

Schwartzmann PV, Ramalho LN, Neder L, Vilar FC, Ayub-Ferreira SM,Romeiro MF, Takayanagui OM, Dos Santos AC, Schmidt A,Figueiredo LT, Arena R, Simoes MV (2017) Zika virus

J. Neurovirol. (2018) 24:273–290 289

meningoencephalitis in an immunocompromised patient. Mayo ClinProc 92:460–466

Sejvar JJ, Leis AA, Stokic DS, Van Gerpen JA, Marfin AA, Webb R,Haddad MB, Tierney BC, Slavinski SA, Polk JL, Dostrow V,Winkelmann M, Petersen LR (2003) Acute flaccid paralysis andWest Nile virus infection. Emerg Infect Dis 9:788–793

Shao Q, Herrlinger S, Yang SL, Lai F, Moore JM, BrindleyMA, Chen JF(2016) Zika virus infection disrupts neurovascular development andresults in postnatal microcephaly with brain damage. Development143:4127–4136

Shrestha B, Gottlieb D, Diamond MS (2003) Infection and injury ofneurons by West Nile encephalitis virus. J Virol 77:13203–13213

Shrestha B, Pinto AK, Green S, Bosch I, Diamond MS (2012) CD8+ Tcells use TRAIL to restrict West Nile virus pathogenesis by control-ling infection in neurons. J Virol 86:8937–8948

Siddharthan V, Wang H, Motter NE, Hall JO, Skinner RD, SkirpstunasRT, Morrey JD (2009) Persistent West Nile virus associated with aneurological sequela in hamsters identified by motor unit numberestimation. J Virol 83:4251–4261

Smith DE, Beckham JD, Tyler KL, Pastula DM (2016) Zika virus diseasefor neurologists. Neurol Clin Pract 6:515–522

Soares CN, Brasil P, Carrera RM, Sequeira P, de Filippis AB,Borges VA, Theophilo F, Ellul MA, Solomon T (2016)Fatal encephalitis associated with Zika virus infection in anadult. J Clin Virol 83:63–65

Solomon T, Kneen R, Dung NM, Khanh VC, Thuy TT, Ha DQ, Day NP,Nisalak A, Vaughn DW, White NJ (1998) Poliomyelitis-like illnessdue to Japanese encephalitis virus. Lancet 351:1094–1097

Solomon IH, Milner DA, Folkerth RD (2016) Neuropathology of Zikavirus infection. J Neuroinfect Dis 7

Sumathy K, Kulkarni B, Gondu RK, Ponnuru SK, Bonguram N, EligetiR, Gadiyaram S, Praturi U, Chougule B, Karunakaran L, Ella KM(2017) Protective efficacy of Zika vaccine in AG129 mouse model.Sci Rep 7:46375

Taylor MP, Enquist LW (2015) Axonal spread of neuroinvasive viralinfections. Trends Microbiol 23:283–288

van den Broek MF, Muller U, Huang S, Aguet M, Zinkernagel RM(1995) Antiviral defense inmice lacking both alpha/beta and gammainterferon receptors. J Virol 69:4792–4796

van den Pol AN, Mao G, Yang Y, Ornaghi S, Davis JN (2017) Zika virustargeting in the developing brain. J Neurosci 37:2161–2175

Wang H, Siddharthan V, Hall JO, Morrey JD (2009) West Nile viruspreferentially transports along motor neuron axons after sciaticnerve injection of hamsters. J Neurovirol 15:293–299

Watson C, Paxinos G, Kayalioglu G, Heise C (2009) Atlas of the mousespinal cord. In: Watson C, Paxinos G, Kayalioglu G (eds) Spinalcord. A Christopher and Dana Reeve Foundation Text and Atlas.Elsevier Ltd, San Diego, pp 308–379

Weger-Lucarelli J, Duggal NK, Bullard-Feibelman K, Veselinovic M,Romo H, Nguyen C, Ruckert C, Brault AC, Bowen RA, StengleinM, Geiss BJ, Ebel GD (2017) Development and characterization ofrecombinant virus generated from a new world Zika virus infectiousclone. J Virol 91

Winkler CW, Myers LM, Woods TA, Carmody AB, Taylor KG, PetersonKE (2017) Lymphocytes have a role in protection, but not in patho-genesis, during La Crosse virus infection in mice. JNeuroinflammation 14:62

Xiao SY, Guzman H, Zhang H, Travassos da Rosa AP, Tesh RB (2001)West Nile virus infection in the golden hamster (Mesocricetusauratus): a model for West Nile encephalitis. Emerg Infect Dis 7:714–721

Zmurko J, Marques RE, Schols D, Verbeken E, Kaptein SJ, Neyts J(2016) The viral polymerase inhibitor 7-Deaza-2 ′-C-methyladenosine is a potent inhibitor of in vitro Zika virus replica-tion and delays disease progression in a robust mouse infectionmodel. PLoS Negl Trop Dis 10:e0004695

Zukor KA, Kent DT, Odelberg SJ (2010) Fluorescent whole-mount meth-od for visualizing three-dimensional relationships in intact andregenerating adult newt spinal cords. Dev Dyn 239:3048–3057

Zukor K, Wang H, Hurst BL, Siddharthan V, Van Wettere A, PilowskyPM, Morrey JD (2017) Phrenic nerve deficits and neurological im-munopathology associated with acute West Nile virus infection inmice and hamsters. J Neurovirol 23(2):186–204

290 J. Neurovirol. (2018) 24:273–290